To The Moon

The Financial Times recently reported that SK Hynix's ADR offering was seven times oversubscribed, with more than 500 institutional investors competing for shares. Even investors such as Leo and Philippe reportedly received smaller allocations because demand far exceeded supply.

$SK hynix(SKHY)$  

$hynix-WI(SKHYV)$  

Yet the most striking detail wasn't the demand.

It was the valuation.

The report noted that SK Hynix's Korean-listed shares were trading at just 4.5x forward earnings.

4.5x. Let that sink in.

One of the most fascinating things about SK Hynix is how consistently the market has refused to rerate the business.

Since 2025, its forward P/E has hovered around just 6x, and now, despite becoming arguably the world's most important AI memory supplier, the valuation has compressed to around 4.5x.

What does that tell us?

It tells us that, despite the stock's enormous gains over the past two years, the market still views SK Hynix as a traditional cyclical memory company rather than a structural growth business.

Almost every dollar of share price appreciation has been driven by EPS growth, not by investors assigning a higher valuation multiple.

That's a remarkable disconnect.

It also means memory investors don't necessarily need multiple expansion to generate attractive returns. If earnings continue compounding, even a business permanently valued at 5–6x earnings can deliver exceptional shareholder returns through EPS growth alone.

The market remains anchored to decades of history, when memory was defined by brutal cycles, chronic oversupply, and collapsing margins. Many investors still assume this cycle will end the same way.

But today's industry looks fundamentally different.

AI is creating an unprecedented need for high-bandwidth memory, supply discipline has improved dramatically, and technology leadership has become more concentrated than ever. The biggest uncertainty remains how quickly China can build competitive capacity. Beyond that, the industry's structure appears significantly healthier than in previous generations.

History also suggests that markets rarely keep exceptional businesses trading at distressed valuations forever.

If SK Hynix continues compounding earnings while trading at just 4.5–6x forward earnings, the math becomes increasingly difficult to ignore. The company could generate enormous free cash flow, aggressively repurchase shares, and still invest for growth. Eventually, such a valuation becomes inconsistent with the economics of the business.

The really interesting question isn't whether earnings can continue growing.

It's whether the market eventually realizes it has been valuing the wrong company.

If investors stop viewing SK Hynix as just another cyclical memory manufacturer and instead recognize it as a critical AI infrastructure company, then multiple expansion could become a second engine of returns alongside earnings growth.

For the past two years, shareholders have benefited almost entirely from EPS growth.

If a genuine rerating ever begins, the next phase could look very different.

This version avoids making a specific prediction (such as "10x from here") while making the central investment thesis stronger: the upside could come from both continued EPS growth and eventual multiple expansion.